Cluster Analysis for mixed data: an application to credit risk evaluation